PROJECT SUMMARY Pediatric solid tumors are a rare and highly heterogeneous collection of cancers. For many subtypes, progress in defining novel therapies has stalled over the last 10-20 years. Indeed, for most of these tumors chemotherapy continues to be the primary form of treatment and targeted therapies are not available. This lack of progress is likely due at least in part to the difficulty in developing clinical trials for individual histologic subtypes given their rarity. An alternative is to develop a robust preclinical testing program. Currently, such programs are limited by the lack of well-credentialed models that incorporate the most advanced technologies for genomic and functional characterization and do to the lack of good models for therapy-resistant and metastatic disease. Here we bring together two PIs with complimentary expertise to develop new approaches for validation and preclinical use of pediatric solid tumor animal models. Our focus is on the use of patient- derived xenografts for three most common histotypes: osteosarcoma, ewing sarcoma and rhabdomyosarcoma. PDX models are particularly well suited for studying highly heterogeneous tumors such as pediatric solid tumors. In Aim 1, we will develop novel advance PDX models that incorporate two key innovations: i) utilization of CRISPR/CAS9 technology for genetic interrogation and ii) autologous and allogeneic approaches for development of PDX models with a human immune system (hu-PDX). In Aim 2, we will develop novel computational tools to assess the similarity of PDX models to their tumor of origin and to evaluate human- mouse and mouse-mouse evolution which may impact clinical relevance. We will work closely with members of the Oncology Models Forum to develop scoring systems to assess this similarity and make these tools widely available to the modeling community. In Aim 3, we will evaluate intratumor heterogeneity during human-mouse and mouse-mouse evolution of PDX models using single cell RNAseq. We expect that the tools and models developed here will be widely applicable to other PDX models and that the specific models we develop will help facilitate preclinical research. We will make all tools and models widely accessible to the research community.